The Genuine Progress Indicator (GPI) is the left-leaning Australia Institute‘s attempt to produce a broader assessment of economic wellbeing than is provided by GDP. At a time when the community faces what the great economist J. R. Hicks referred to as ‘the diminishing marginal significance of economics’ broader measures of economic welfare are increasingly needed.
And when we look at the broader measure we find some of the Australia Institute’s worst fears confirmed. Ours is a case of joyless growth. Here’s the way Geoff Davies put it in a recent piece for the Centre for Policy Development which took my eye.
‘The GPI uses a proper balance sheet, crediting the good, debiting the bad and calculating the total. The GPI also includes environmental costs such as the loss of soil and forests, oil depletion and climate change. The result of this more sensible accounting is salutary — for many Western countries the GPI increased steadily through the 1950s and 1960s but has been pretty stagnant since the 1970s. In other words we have been busier and have spent more, but the net benefit is hardly any greater than it was thirty years ago. This gives us a pretty clear message that we’re not using our time well.’
But however much one might sympathise with a cause — as I do with the basic idea behind the GPI — a lack of hard headedness in execution leaves me not only disappointed but wondering whether the exercise isn’t counter-productive. The GPI is unpersuasive even to someone sympathetic to its basic intentions, because a closer look demonstrates that it is built to illustrate a range of complaints about the costs of economic growth that became fashionable in the 1970s. That need not be a problem — for most of the complaints are legitimate enough. But while it takes most opportunities to deduct some of the less attractive things about recent economic growth from its measure of economic wellbeing, the GPI pays almost no attention to the positives that have come our way as well.
In this and a subsequent article to be published next week I offer some clarifications of the conceptual background behind GDP. I then demonstrate the GPI’s biases. Towards the end of this article I show how methods of calculation appear to bias the results. Next week I will look more broadly at what has been measured and what has been left out.
National income accounting matured from the late 1930s on, underpinned by the newfound importance of aggregate measures of economic activity like consumption, investment and saving in the new discipline of macroeconomics. Tracking measures such as the total goods and services produced in an economy (GDP) were obviously important to the new goal of managing the economic cycle.
It’s worth stressing that national income accounting is just that — accounting. And accounting makes simplifications which enable quantities to be compared. A single person or a family might aggregate their own earnings or wealth in a year, while being fully aware of the limitations of this as a measure of their economic welfare, let alone their welfare more generally.
This was the origin of GDP and similar measures, and the economists and statisticians that constructed the measures knew (as do economists trained today) that they are not direct measures of welfare, but rather the process of a relatively coherent set of accounting simplifications or conventions. However, GDP came to be associated more and more, by the media, by politicians and by more mediocre and slapdash economists as something akin to a direct measure of economic welfare.
It was also useful fodder for our habitual (and intensifying?) tendency to compare ourselves to other countries as if we were in competition with them, when we should be minding our own business – quite literally in this case – and leading the lives we want to lead.
In the age of the ‘factoid’, when you need a simple number to communicate anything through our media, simply insisting on the limitations of the GDP — insisting on its necessary limitations as a piece of aggregate accounting — doesn’t cut the mustard. Your qualifications fade into insignificance as the number makes its way through the media as a sound bite.
This, and a genuine belief in the worth of trying for some better measure of aggregate economic welfare led to a range of proposals to supplement the GDP so as to produce better estimates of economic wellbeing.
It can’t do it perfectly of course, but the GPI begins with a better measure of welfare than GDP. As Adam Smith insisted all those years ago, however much businesspeople might like us to forget it, we produce to consume, not vice versa.
So GPI begins with a simple accounting measure of national economic consumption rather than production. It then corrects that figure by adding and subtracting from it. It subtracts from it where consumption becomes more unequal — a subjective value judgement but one that’s fine by me. Likewise with the deduction of 50 per cent of the cost of advertising.
And what could be more sensible than subtracting the costs of resource depletion or an estimate of the costs of pollution? While we’re about it shouldn’t we add the value of home-based production? Certainly omitting them creates a suite of paradoxes — for instance that one increases the GDP by putting one’s children into childcare, but not by looking after them oneself. The GPI does all this — or at least attempts to.
A serious rebalancing of a single index of economic welfare will contain inevitable simplifications with which one can take issue. But what if every single change that was made took the index in the direction of the prejudices of its authors?
Of the twenty columns subtracting value from the GPI all subtract far more in 2000 than they do in 1950. So every one depresses GPI growth compared with GDP growth over the period. Only one column — the costs of ozone depletion — was clearly trending down in the mid 1990s when the Australian GPI was first developed. By 2000 a second column was slowly trending down. In each case, because the amount they are subtracting from GPI is reducing, each is now making a modest positive contribution to GPI growth.
Let’s take a closer look — first at how some of the numbers are calculated, and then at some columns that might add to the GPI and add to measured growth — but have not been included.
The GPI tells us that the costs of transport accidents have been bouncing along with no discernable trend in a band between $4.5 billion and $6 billion per annum from the early 1970s. But the Australian Bureau of Statistics says that the road toll reached its high point in 1970 with 3,798 deaths. This equates to 30.4 per 100,000 people or 8 deaths per 10,000 cars. By 2003 the road toll had fallen to 1,633 people at a rate of 8.2 deaths per 100,000 people or 1.2 deaths per car (link here).
Somehow this spectacular improvement doesn’t make it into the GPI. The GPI methodology is simplified — as it must be. Simplifications are inevitable in any such exercise, but doubly so given the limited resources available for the GPI project. So reported transport deaths are taken and indexed by GDP.
Now the number of deaths are the product of the death rate and our population. So population growth has already been counted. So if GDP numbers are the appropriate figure by which to index costs (which is itself doubtful, but we will not consider this further here), the appropriate index is GDP per capita. The GPI’s use of GDP thus seems to effectively double count population growth.
Over a few decades this makes quite a difference. Whether it’s the whole explanation for the disparity between the GPI and the ABS versions of reality I don’t know. But the dose is repeated, and indeed exacerbated in the next column.
Industrial accidents have also been steadily falling for a substantial period. The methodology in the GPI simply takes a figure for such costs in 1992-93 and indexes it by the size of the labour market and by GDP. Thus in addition to removing the effect of improved accident and injury rates, population growth again gets double counted, once as labour market growth and again as that part of the GDP attributable to population growth.
Thanks to Clive Hamilton, Richard Denniss, and Ian Castles for helpful comments on earlier drafts. Mistakes of fact or interpretation that remain are those of the author.
